has anyone attempted to use a program like novoalign to better estimate the actual error rate of illumina sequencing protocol? i saw some discussion here a while ago.
i saw an estimate here although i'm not clear what the source was: ~180 errors per 100 of 2x36bp reads
i've seen some work this year on minimizing the effect of an intrinsic error rate on mappability using parent-child clustering but i haven't seen any new data on what the likely true error rate is. I would think that a robust aligner like NovoAlign - since it can align with up to 7 mismatches in SE reads (in my hands this amounts to >60% of all reads unmappable by MAQ) - would provide better estimate of error rate than the illumina method...which i'm unsure what that method actually is but my experience with SOLiD and Illumina make me think that it's an underestimate.
Thanks for conversing!
der. (Todd)
i saw an estimate here although i'm not clear what the source was: ~180 errors per 100 of 2x36bp reads
i've seen some work this year on minimizing the effect of an intrinsic error rate on mappability using parent-child clustering but i haven't seen any new data on what the likely true error rate is. I would think that a robust aligner like NovoAlign - since it can align with up to 7 mismatches in SE reads (in my hands this amounts to >60% of all reads unmappable by MAQ) - would provide better estimate of error rate than the illumina method...which i'm unsure what that method actually is but my experience with SOLiD and Illumina make me think that it's an underestimate.
Thanks for conversing!
der. (Todd)